Running and testing

​

Run the whole workflow

Click Run in the top right (or press Shift + R). While a run is in progress the button becomes Stop; if several runs are active it becomes a Stop (N) dropdown where you can stop one run or Stop all. If a workflow only starts on its own, with triggers like an incoming email, a new row, a schedule, or a webhook and no manual entry point, pressing Run opens a popup that explains it runs automatically. The popup lists what each trigger watches and the address or URL that fires it. From there you can Add a Run step to add a manual entry point, Open config for any trigger, or Run anyway to do a single manual run. See Triggers overview. When you Run anyway on a workflow whose webhook or inbound-email trigger has already fired at least once, the trigger replays its last received event so the rest of the flow has real data to work with. A trigger that has never fired runs as before.

​

Before running

If any step still needs setup, Run opens a popup that walks you through them one at a time, with a progress bar you can click to jump around. A pill at the top counts what is left (“2 nodes need setup”) and turns into Ready to run once you clear them. A step appears here when it is missing a required field, has no action selected, has no account connected, or is a tool provider with no operations allowed yet. You fix each one in the popup itself:

• Missing fields get the same editors as the config panel, including dropdowns that load their options from the connected account.
• No action selected gets the node’s action picker. Until an action is picked, which fields the node needs is unknown, so this comes first.
• No account gets the node’s credentials block. Connecting through OAuth opens the provider window with the popup still up, and the block flips to done when the credential lands.
• A tool provider with nothing allowed gets its operation list, so you can tick the operations the agent may call.

Requirements stay listed once you satisfy them, marked as done, so a block never disappears mid-edit. The run action is always available: it reads Run anyway while something is unfilled and Run when everything is set. Open config on any step takes you to that node on the canvas instead. This check also runs before a single node run and before Run from here, over the nodes those runs actually touch, which includes upstream nodes whose data is referenced and any tool providers wired into an agent that runs. On mobile, Run starts the run without the popup.

​

Choose where a run starts

When a workflow has more than one entry point, the popup also asks which of them to start from. Each row names the branch it begins, such as “Then Slack, Gmail” or “Uses Telegram”, and you untick the ones to leave out. Untick everything and Run is disabled. With more than two entry points, Select all and Clear appear. Interface blocks that start nothing, such as a standalone app or a display block, are not listed. They render from their own config in the Interface tab, so there is nothing to run. An interface block that does feed other nodes, like a form leading into Slack, is still listed, because the branch behind it runs only when you tick it. An agent entry point gets a box for its opening message, prefilled from the agent’s Message field, so you can adjust what it does on this run without editing the node. When the agent is the only entry point you run, its answers arrive in the chat, which opens as soon as the run starts. With another branch ticked too, the run stays on the canvas and the agent’s answer lands on its node. This screen appears when there is a choice to make or an agent message to write. A workflow with one non-agent entry point runs straight off the button. Workflows that only start on their own show the trigger popup described above instead.

​

Run results

When a run finishes, a results popup shows what each node produced. A left rail lists the nodes that ran, each with a status icon, and the main panel shows the selected node’s output. For an AI agent, you see its final response along with the tool calls it made. Use Open config on any node to jump straight to its settings. The popup opens on its own after a full run, a Run from here run that starts partway through the flow, and runs started by a trigger such as a schedule or a webhook. To stop it from opening automatically, click Don’t show again; turn it back on under Settings in the Popups section. These popups do not appear on mobile. The Runs pill on the top left of the canvas is always available, even before this session’s first run. It shows the latest run’s status and how long ago it ran, and clicking it reopens the results popup. Inside the popup, a run switcher lets you load any earlier run and review its outputs, scrolling back through your run history without leaving the canvas.

​

Run a single node

Testing one node at a time is usually faster than full runs: If the node reads from upstream nodes that haven’t produced output yet, the Run button is disabled (“Missing input data from previous nodes”). Run the upstream nodes first, or mock their outputs.

​

Inspecting input and output

With a node selected, toggle Output (Y) to see what it produced, in field, JSON, or table view, and Input (X) to see the data arriving from connected nodes. Every value in these panels is draggable into downstream config fields as a reference.

​

Mocked outputs

Click Mock (the pin icon) in the Output panel to pin the currently displayed output as the node’s mocked output. The node then shows a MOCK badge, and downstream nodes use the pinned data without the node re-running, so you can build and test the rest of the workflow against a stable sample. Click Clear Mock (or press M with the node selected) to go back to live output. Save stores the current output under a name so you can reload it as mock data later (3 saved outputs per type on Free, 10 on Plus, unlimited on Pro). You can also pin a node’s output without opening the Output panel: hover the node and click the pin button on its top-left edge. It stays highlighted while the output is pinned, and you click it again to clear. The button is available once the node has produced output.

​

Execution states on nodes

Nodes reflect the run live on the canvas: a node animates while it is running, settles when completed, and shows error styling when it fails. Disabled nodes are skipped entirely. Toggle a node off with D, or hover it and click the disable button on its bottom-left edge; a disabled node greys out and the button turns amber until you enable it again.

​

Next steps